• Amina Saleh Omar
• Mvurya Mgala
• Fullgence Mwakondo

This study presents a comprehensive unified model for the early detection and support of dysgraphia in primary school learners by integrating deep learning, handwriting feature analysis, and pedagogically grounded intervention strategies. Utilizing DenseNet121 for feature extraction and a tuned Artificial Neural Network (ANN) for classification, the proposed model achieved high diagnostic accuracy (95.56%) and strong generalization to real-world school data (92.3%). Quantitative handwriting features were extracted from confirmed dysgraphia cases using OpenCV functions in python and clustered using KMeans to stratify severity into Mild, Moderate, and Severe profiles. These profiles informed tailored visual scaffolding strategies derived from a six-level cognitive framework, supporting instructional planning. The model was validated through both benchmark and field testing, demonstrating robustness, interpretability, and potential for scalable integration into educational settings. Although real-time deployment was not implemented, strategy recommendations were made for future adoption in tablet-based learning environments. The findings highlight the utility of combining artificial intelligence with cognitive scaffolding for inclusive and data-driven handwriting interventions.

Dysgraphia Detection, Deep Learning, Handwriting Analysis, Scaffolding Strategies, Educational Technology.

"A UNIFIED AI MODEL FOR SUPPORTING DYSGRAPHIA LEARNERS USING VISUAL SCAFFOLDING TECHNIQUES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 7, page no.a674-a691, July-2025, Available :http://www.jetir.org/papers/JETIR2507072.pdf

"A UNIFIED AI MODEL FOR SUPPORTING DYSGRAPHIA LEARNERS USING VISUAL SCAFFOLDING TECHNIQUES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 7, page no. ppa674-a691, July-2025, Available at : http://www.jetir.org/papers/JETIR2507072.pdf